https://support.terra.bio/hc/en-us/articles/360041068771

2020.08.23.20178236v1.full.sars.kraken2.terra.pdf

这个文章里面提到 使用kraken2检测其他病毒 主要用于排除新冠和其他病毒的交叉感染

We used Kraken2 (46) to identify other viral taxa present in NP swab samples from COVID
positive patients, excluding those removed by filters i and ii described above. To do so, we ran
the classify_single workflow on all reads from all samples (with
kraken2_db_tgz=”gs://pathogen-public-dbs/v1/kraken2-broad-20200505.tar.zst”,
krona_taxonomy_db_kraken2_tgz=”gs://pathogen-public-dbs/v1/krona.taxonomy-20200505.tab.
zst”, ncbi_taxdump_tgz=”gs://pathogen-public-dbs/v1/taxdump-20200505.tar.gz”,
trim_clip_db=”gs://pathogen-public-dbs/v0/contaminants.clip_db.fasta”,
spikein_db=”gs://pathogen-public-dbs/v0/ERCC_96_nopolyA.fasta”). Our kraken2 database was

Terra的这个流程是针对illumina二代的情况
https://app.terra.bio/#workspaces/pathogen-genomic-surveillance/COVID-19_Broad_Viral_NGS

• basecalling
  Nanopore Guppy base caller (v3.4.4) tool
  “flow cell = FLO-MIN106” and “kit = SQK-RNA002”.

• 质控
  PycoQC (v2.5.0.21) software

• 过滤
  NanoFilt (v2.7.0)
  minimum read length ≥500 nt and read quality ≥8.

• 去宿主和其他微生物(去污染)
  去除人GRCh38 (hg38) fungal and bacterial genome
  minimap2 (v2.17–r941)

• 提取新冠病毒
  samtools (v1.7) view unmapped reads and reads with mapping quality lower than 10

• 对齐 call 突变
  minimap2
  BCFtools mpileup\call
  使用Integrative Genomic Viewer (IGV) (v2.8.2) 查看突变

这个文章对新冠的UTR部分的突变做了分析

这个文章的附件有一张表 里面有SRR号 序列数很多 对我们来说需要支持这种批量下载数据的情况

这个文章里面对武汉新冠当时的情况做了一个介绍 5个样本中国疾控处理、4个样本华大处理的
华大用bwa和hg19人的基因组对比 去除了宿主 然后和NCBI的冠状病毒（具体是哪个序列还不知道）做了对齐 然后使用SPAdes做了一个一致性序列
中国疾控用的是 CLCBio （就是我研究的竞品 CLC workbench）software version 11.0.1 was used for de novo assembly, variant calling, and alignment
有了这些组装的结果之后，就可以做进化分析（ phylogenetic analysis）

该文章对covid-19 NGS实验和生信做了综述 里面引用了一些文章还是有价值的：
1.https://pubmed.ncbi.nlm.nih.gov/29154853/
Standards and Guidelines for Validating Next-Generation Sequencing Bioinformatics Pipelines: A Joint Recommendation of the Association for Molecular Pathology and the College of American Pathologists
这个文章里面总结了一个专家共识，提出了17条临床NGS生物信息学管道验证的最佳实践共识建议

属于解释
Terminology
Description
BAM
Compressed (binary) format of a SAM file intended for faster random access (search) of aligned and unaligned sequences and its related metadata. Compression enables smaller file size and storage efficiency which makes it popular over the SAM format. This is a default output of many alignment and post-alignment softwares used in bioinformatics pipeline and commonly used as an input by many variant callers*.
FASTQ
It is a de facto, human readable, file format that stores nucleotide sequences and corresponding quality (PHRED) scores for each nucleotide as an ASCII encoded character.4 This is commonly used for storing unaligned short sequence reads after the steps of base calling and is a typical starting point for NGS bioinformatics pipeline.
PHRED Score
It is a per base (nucleotide) quality score that is defined as an estimated probability for a called base to be incorrect (erroneous call). Mathematically, it is expressed as4

where Q is Phred quality score, Pe is the probability for an erroneous base call. The Pe is typically generated by the base calling software which is sequence instrument specific. Therefore, Q values in isolation cannot be used to compare sequence quality across different sequencing platforms.
SAM
Stands for Sequence Alignment/Map format. It is a human readable (text file) file format specification for storing information on aligned sequence. This is a default output of many alignment softwares used in bioinformatics pipeline. Given the large file size and slower random access, BAM format is preferred for routine bioinformatics data processing. This format is helpful for technical troubleshooting when manual review of the stored information is necessary*.
Variant - horizontally complex
When two or more sequence alterations are present on the same read in close proximity such that they may represent a single complex variant. These variants are frequently represented as deletion-insertions and may result in ambiguous sequence description or HGVS nomenclature.
Variant - Left-aligned
If there are multiple potential VCF entries of the same allele length that represent the same variant, then left-alignment refers to the VCF entry with the smallest base position. The base position is typically represented in genomic coordinate for a given primary assembly (eg, GRCh38) and represents the most 5’ position.5
Variant - Normalized
A normalized variant must be parsimonious as well as left-aligned.5
Variant - Parsimony
If there are more than one way to represent the same variant in a VCF file, parsimony refers to the representation with the shortest possible allele length5 (positive and non-zero length).
Variant - vertically complex
A vertically complex variant occurs when three or more alleles are represented by different sequence reads, typically with or uncommonly without a reference (normal) allele, at the same genomic coordinate or set of coordinates.
VCF
Variant Call Format is a versioned, text-file (human readable) specification for storing sequence variant calls. The file contains meta-information containing various details of the variant calling process and definition of headers and format tags, a header line and data lines. Each data line represents a sequence variant defined using a combination of chromosome, position, reference allele, and alternate allele†.

MUSCLE v3.8.1551 by Robert C. Edgar

http://www.drive5.com/muscle
This software is donated to the public domain.
Please cite: Edgar, R.C. Nucleic Acids Res 32(5), 1792-97.

nCov19.muscle.in 2 seqs, lengths min 29903, max 29903, avg 29903
00:00:00 20 MB(-2%) Iter 1 100.00% K-mer dist pass 1
00:00:00 20 MB(-2%) Iter 1 100.00% K-mer dist pass 2
00:01:16 1004 MB(-116%) Iter 1 100.00% Align node
00:01:16 1004 MB(-116%) Iter 1 100.00% Root alignment

详细过程分析

1.nanopolish index -s guppy_fastq/sequencing_summary.txt -d /mnt/sdf1/sars_2019/ guppy_fastq/all_sars19.fastq

nanopolish需要使用原始数据，index命令会将原始数据和guppy产生的reads关联起来

2.minimap2 -a -x map-ont -t 8 /home/bioinfo/artic_ncov_2019/artic-ncov2019/primer_schemes//nCoV-2019/V3/nCoV-2019.reference.fasta guppy_fastq/all_sars19.fastq | samtools view -bS -F 4 - | samtools sort -o nCov19.sorted.bam -
samtools index nCov19.sorted.bam
将序列和参考序列进行比对　生成bam文件 此步骤也可以使用bwa mem进行比对

3.align_trim --start --normalise 100 /home/bioinfo/artic_ncov_2019/artic-ncov2019/primer_schemes//nCoV-2019/V3/nCoV-2019.scheme.bed --report nCov19.alignreport.txt < nCov19.sorted.bam 2> nCov19.alignreport.er | samtools sort -T nCov19 - -o nCov19.trimmed.rg.sorted.bam
目的：这一步严重依赖primer的相关知识　相关论文和工具可以从下面链接查到
https://primalscheme.com/
The purpose of alignment post-processing is:
assign each read alignment to a derived amplicon
using the derived amplicon, assign each read a read group based on the primer pool
softmask read alignments within their derived amplicon

4.突变分析 medaka-longshot　流程　在这个流程中Longshot is used in the ARTIC pipeline simply to annotate the VCF with various statistics (like read support for ALTs). The newest medaka v1.0.0 has a tool to perform this operation:可以被medaka tools annotate --help　取代
medaka consensus
medaka variant or snps (if pipeline has been told not to detect INDELS via --no-indels)
medaka tools annotate (if --no-longshot has been selected)
longshot (if --no-longshot not selected)

另外一个突变流程是nanopolish　但是medaka宣称　50X faster than Nanopolish (and can run on GPUs).

5.生成一致性序列

bgzip barcode$i.vcf
tabix -p vcf barcode$i.vcf.gz

longshot -P 0 -F -A --no_haps --bam barcode$i.bam --ref ~/ref/h1n1/ref.fasta --out barcode$i.longshot.vcf --potential_variants barcode$i.vcf.gz
artic_vcf_filter --longshot barcode$i.longshot.vcf barcode$i.pass.vcf barcode$i.fail.vcf
artic_make_depth_mask ~/ref/h1n1/ref.fasta barcode$i.bam barcode$i.coverage_mask.txt

bgzip -f barcode$i.pass.vcf
tabix -p vcf barcode$i.pass.vcf.gz

artic_mask ~/ref/h1n1/ref.fasta barcode$i.coverage_mask.txt barcode$i.fail.vcf barcode$i.preconsensus.fasta
bcftools consensus -f barcode$i.preconsensus.fasta barcode$i.pass.vcf.gz -m barcode$i.coverage_mask.txt -o barcode$i.consensus.fasta

https://github.com/nanoporetech/medaka/issues/149
longshot只是做了一些统计

https://artic.readthedocs.io/en/latest/release-notes/
Longshot added to Medaka to permit filter VCF on depth
Longshot似乎只是做了一些过滤

https://bedtools.readthedocs.io/en/latest/content/tools/maskfasta.html
咱们的过程没有用到 medaka tools annotate 步骤 这个步骤和longshot是二选一的过程
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7480024/